Last request for testers for a common tournament :-)

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 17 July 2004 17:30:38:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.
There were a lot of new releases over the past weeks, a lot of engines improved tremendously and this will also be tested here.
For the middle class and weaker tournament there are also testers needed. Authore who test will have their engines included, but their respective engine will be tested by others for neutrality.
Hoping for more assistance.
Heinz

[List of testers up to now]

Geschrieben von:/Posted by: List of testers up to now Heinz van Kempen at 17 July 2004 17:39:09:
Als Antwort auf:/In reply to: Last request for testers for a common tournament geschrieben von:/posted by: Heinz van Kempen at 17 July 2004 17:30:38:

Hi ,
for Master Class (tournament A) we have up to now:

Volker Pittlik
Olivier Deville
Igor Gorelikov
Slobodan R. Stojanovic
Christian Koch
Roger Brown
Heinz van Kempen
for higher class (tournament B):
Thomas Mayer
Volker Anuss
Olivier Deville (second computer)
Heinz van Kempen (second computer)
More testers would make possible, more games, more engines, more time for each games, more reliable results......
For interest please post here or write to hvankempen@web.de
Best Regards
Heinz

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 17 July 2004 17:46:19:
Als Antwort auf:/In reply to: Re: Last request for testers for a common tournament geschrieben von:/posted by: List of testers up to now Heinz van Kempen at 17 July 2004 17:39:09:

[Bryan Hofmann]

Geschrieben von:/Posted by: Bryan Hofmann at 18 July 2004 15:04:22:
Als Antwort auf:/In reply to: Last request for testers for a common tournament geschrieben von:/posted by: Heinz van Kempen at 17 July 2004 17:30:38:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.

Here is a small test I did with Crafty dealing with the Learn off vs on. I ran a 5/2 match Crafty 19.15 vs Riffian 1.0.5 using the Nunn I & II openings and mirroring them so both engines played each opening as White & Black. As you can see there is a 65 point difference in strength. Since there were no books used in this match this represents the strength difference of position learning only.

Crafty 19.15 Learn Off - Ruffian 1.0.5 : 17.5/60 9-34-17 (==1=0000==000==00==110000000==010=1001010=01=00=0=0000=00100)
Elo : -154
Margins :
68 % : (+ 47,- 33) -> [-187,-107]
95 % : (+ 90,- 70) -> [-224, -64]
99.7 % : (+131,-112) -> [-266, -23]

Crafty 19.15 Learn On - Ruffian 1.0.5 : 22.5/60 13-28-19 (=0==0100==01==01==010=000000===101=011110=01==1=0=0000010000)
Elo : -89
Margins :
68 % : (+ 39,- 37) -> [-126, -50]
95 % : (+ 76,- 78) -> [-166, -13]
99.7 % : (+113,-123) -> [-211, 24]

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 18 July 2004 15:13:41:
Als Antwort auf:/In reply to: Re: Last request for testers for a common tournament geschrieben von:/posted by: Bryan Hofmann at 18 July 2004 15:04:22:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.

Here is a small test I did with Crafty dealing with the Learn off vs on. I ran a 5/2 match Crafty 19.15 vs Riffian 1.0.5 using the Nunn I & II openings and mirroring them so both engines played each opening as White & Black. As you can see there is a 65 point difference in strength. Since there were no books used in this match this represents the strength difference of position learning only.

Crafty 19.15 Learn Off - Ruffian 1.0.5 : 17.5/60 9-34-17 (==1=0000==000==00==110000000==010=1001010=01=00=0=0000=00100)
Elo : -154
Margins :
68 % : (+ 47,- 33) -> [-187,-107]
95 % : (+ 90,- 70) -> [-224, -64]
99.7 % : (+131,-112) -> [-266, -23]

Crafty 19.15 Learn On - Ruffian 1.0.5 : 22.5/60 13-28-19 (=0==0100==01==01==010=000000===101=011110=01==1=0=0000010000)
Elo : -89
Margins :
68 % : (+ 39,- 37) -> [-126, -50]
95 % : (+ 76,- 78) -> [-166, -13]
99.7 % : (+113,-123) -> [-211, 24]

[sorry, wrong button Heinz]

Geschrieben von:/Posted by: sorry, wrong button Heinz van Kempen at 18 July 2004 15:21:12:
Als Antwort auf:/In reply to: Re: Last request for testers for a common tournament geschrieben von:/posted by: Bryan Hofmann at 18 July 2004 15:04:22:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.

Here is a small test I did with Crafty dealing with the Learn off vs on. I ran a 5/2 match Crafty 19.15 vs Riffian 1.0.5 using the Nunn I & II openings and mirroring them so both engines played each opening as White & Black. As you can see there is a 65 point difference in strength. Since there were no books used in this match this represents the strength difference of position learning only.

Crafty 19.15 Learn Off - Ruffian 1.0.5 : 17.5/60 9-34-17 (==1=0000==000==00==110000000==010=1001010=01=00=0=0000=00100)
Elo : -154
Margins :
68 % : (+ 47,- 33) -> [-187,-107]
95 % : (+ 90,- 70) -> [-224, -64]
99.7 % : (+131,-112) -> [-266, -23]

Crafty 19.15 Learn On - Ruffian 1.0.5 : 22.5/60 13-28-19 (=0==0100==01==01==010=000000===101=011110=01==1=0=0000010000)
Elo : -89
Margins :
68 % : (+ 39,- 37) -> [-126, -50]
95 % : (+ 76,- 78) -> [-166, -13]
99.7 % : (+113,-123) -> [-211, 24]

Hello Bryan,
I got the first votes from our testers and it might possibly be a close decision if book learning and position learning should be on or off.
In case that a majority votes for learning off....
What would you think if we opt for the following: for the first bunch of engines disabling this things for Crafty 19.15. After completing the first stage and beginning with the gauntlets adding more engines another Crafty 19.15, playing all its games on one fast computer with all those things enabled and then look if we can underline your results after a lot of games?
Best Regards
Heinz

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 18 July 2004 15:33:23:
Als Antwort auf:/In reply to: Re: Last request for testers for a common tournament geschrieben von:/posted by: Heinz van Kempen at 18 July 2004 15:13:41:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.

Here is a small test I did with Crafty dealing with the Learn off vs on. I ran a 5/2 match Crafty 19.15 vs Riffian 1.0.5 using the Nunn I & II openings and mirroring them so both engines played each opening as White & Black. As you can see there is a 65 point difference in strength. Since there were no books used in this match this represents the strength difference of position learning only.

I am surprised to read it.
I assumed that position learning is only relevant after you get the same position that you got in previous game with position learning on so position learning can help only if Crafty has previous experience with position learning of the same position and in the nunn match crafty can have previous experience only in case that it played the same position in the past.
If I understand correctly in your experiment it played only one nunn match with positional learning on so based on my understanding positional learning could not help.
Uri

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 18 July 2004 15:48:54:
Als Antwort auf:/In reply to: Re: Last request for testers for a common tournament geschrieben von:/posted by: Bryan Hofmann at 18 July 2004 15:04:22:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.

Here is a small test I did with Crafty dealing with the Learn off vs on. I ran a 5/2 match Crafty 19.15 vs Riffian 1.0.5 using the Nunn I & II openings and mirroring them so both engines played each opening as White & Black. As you can see there is a 65 point difference in strength. Since there were no books used in this match this represents the strength difference of position learning only.

Crafty 19.15 Learn Off - Ruffian 1.0.5 : 17.5/60 9-34-17 (==1=0000==000==00==110000000==010=1001010=01=00=0=0000=00100)

Interesting to note that Crafty19.11 scored 13.5/40
against Ruffian1.0.5 in Heinz tests if I counted correctly(40 games of the nunn match 4+2) and 13.5/40 is better than 17.5/60
Uri

[Volker Boehm]

Geschrieben von:/Posted by: Volker Boehm at 18 July 2004 15:53:04:
Als Antwort auf:/In reply to: Last request for testers for a common tournament geschrieben von:/posted by: Heinz van Kempen at 17 July 2004 17:30:38:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.
There were a lot of new releases over the past weeks, a lot of engines improved tremendously and this will also be tested here.
For the middle class and weaker tournament there are also testers needed. Authore who test will have their engines included, but their respective engine will be tested by others for neutrality.
Hoping for more assistance.
Heinz

Hi,
are we able to send you a new version of spike for this tournament, if yes until which date? We should have enough improvements to build a Spike 0.6.
Volker

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 18 July 2004 15:57:35:
Als Antwort auf:/In reply to: Re: Last request for testers for a common tournament geschrieben von:/posted by: Volker Boehm at 18 July 2004 15:53:04:

Hi,
are we able to send you a new version of spike for this tournament, if yes until which date? We should have enough improvements to build a Spike 0.6.
Volker

Hello Volker,
in case that the majority of testers agree to use also not released versions for the sake of the development of engines (and I can underline with a lot of games that Spike is developping like hell) you can do that. I will give this for discussion and voting, if testers do not want to reply in the forum.
Another condition would of course be that you have to send this Spike version not only to me, but to all the testers.
Best Regards
Heinz

[Bryan Hofmann]

Geschrieben von:/Posted by: Bryan Hofmann at 18 July 2004 16:40:08:
Als Antwort auf:/In reply to: sorry,, wrong button for the last message geschrieben von:/posted by: sorry, wrong button Heinz van Kempen at 18 July 2004 15:21:12:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.

Here is a small test I did with Crafty dealing with the Learn off vs on. I ran a 5/2 match Crafty 19.15 vs Riffian 1.0.5 using the Nunn I & II openings and mirroring them so both engines played each opening as White & Black. As you can see there is a 65 point difference in strength. Since there were no books used in this match this represents the strength difference of position learning only.

Crafty 19.15 Learn Off - Ruffian 1.0.5 : 17.5/60 9-34-17 (==1=0000==000==00==110000000==010=1001010=01=00=0=0000=00100)
Elo : -154
Margins :
68 % : (+ 47,- 33) -> [-187,-107]
95 % : (+ 90,- 70) -> [-224, -64]
99.7 % : (+131,-112) -> [-266, -23]

Crafty 19.15 Learn On - Ruffian 1.0.5 : 22.5/60 13-28-19 (=0==0100==01==01==010=000000===101=011110=01==1=0=0000010000)
Elo : -89
Margins :
68 % : (+ 39,- 37) -> [-126, -50]
95 % : (+ 76,- 78) -> [-166, -13]
99.7 % : (+113,-123) -> [-211, 24]

Hello Bryan,
I got the first votes from our testers and it might possibly be a close decision if book learning and position learning should be on or off.
In case that a majority votes for learning off....
What would you think if we opt for the following: for the first bunch of engines disabling this things for Crafty 19.15. After completing the first stage and beginning with the gauntlets adding more engines another Crafty 19.15, playing all its games on one fast computer with all those things enabled and then look if we can underline your results after a lot of games?
Best Regards
Heinz

My opinion is still the same nothing should be disabled in any engine. Stating that it is unfair that Crafty has the ability to learn and would have and advantage over other engines or it that earlier engines have it better then later engines that play against Crafty just don't hold water. The learn fuction is a integral part of Crafty and disabling it will not represent a true picture of it strength. AI is indeed the future of chess and this is but one of the examples that any other engine Author can/does use, and if they choose not to use it that's their problem.

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 18 July 2004 16:54:20:
Als Antwort auf:/In reply to: Re: sorry,, wrong button for the last message geschrieben von:/posted by: Bryan Hofmann at 18 July 2004 16:40:08:

My opinion is still the same nothing should be disabled in any engine. Stating that it is unfair that Crafty has the ability to learn and would have and advantage over other engines or it that earlier engines have it better then later engines that play against Crafty just don't hold water. The learn fuction is a integral part of Crafty and disabling it will not represent a true picture of it strength. AI is indeed the future of chess and this is but one of the examples that any other engine Author can/does use, and if they choose not to use it that's their problem.

Hello Bryan,
I understand your point of view, it was just a try to conciliate opinions here a bit. Votes so far are two votes against learning, three votes in favour of learning allowed. There are still some more votes to come and testers have time until Monday night at most to decide.
What concerns time control a majority up to now favours 40 moves in 40 minutes for each engine. A lot of different engines received points and it seems that it is not only absolute strength that counts for most testers, but also fast progresses this year, playing style and other factors, for example guessing or knowing, who might be stronger with more time.
This is all no secret for the other testers, as all received the votes from all so far.
More I do not want to tell at this time. It is a lot of suspense.
Best Regards
Heinz

[Pedro Eckmann]

Geschrieben von:/Posted by: Pedro Eckmann at 18. July 2004 17:04:
Als Antwort auf:/In reply to: Last request for testers for a common tournament geschrieben von:/posted by: Heinz van Kempen at 17 July 2004 17:30:38:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.
There were a lot of new releases over the past weeks, a lot of engines improved tremendously and this will also be tested here.
For the middle class and weaker tournament there are also testers needed. Authore who test will have their engines included, but their respective engine will be tested by others for neutrality.
Hoping for more assistance.
Heinz

I would be glad to help for such a noble cause. This is what I got:
Hardware: Intel(R) Pentium(R) 4 CPU 2.53GHz
Betriebssystem: Microsoft Windows XP Build 2600 Service Pack 1
Let me know if I can contribute.
Best Regards
Pedro

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 18 July 2004 17:34:43:
Als Antwort auf:/In reply to: Re: sorry,, wrong button for the last message geschrieben von:/posted by: Heinz van Kempen at 18 July 2004 16:54:20:

My opinion is still the same nothing should be disabled in any engine. Stating that it is unfair that Crafty has the ability to learn and would have and advantage over other engines or it that earlier engines have it better then later engines that play against Crafty just don't hold water. The learn fuction is a integral part of Crafty and disabling it will not represent a true picture of it strength. AI is indeed the future of chess and this is but one of the examples that any other engine Author can/does use, and if they choose not to use it that's their problem.

Hello Bryan,
I understand your point of view, it was just a try to conciliate opinions here a bit. Votes so far are two votes against learning, three votes in favour of learning allowed. There are still some more votes to come and testers have time until Monday night at most to decide.
What concerns time control a majority up to now favours 40 moves in 40 minutes for each engine. A lot of different engines received points and it seems that it is not only absolute strength that counts for most testers, but also fast progresses this year, playing style and other factors, for example guessing or knowing, who might be stronger with more time.
This is all no secret for the other testers, as all received the votes from all so far.
More I do not want to tell at this time. It is a lot of suspense.
Best Regards
Heinz

The question is if amateur programmers are really interested in more long time control games(there are already tournaments like Leo's tournament and the infinite loop).
I think that the time management problem may be more important and programmers may find bugs in their time management if they find that their engine perform significantly worse in 4 minutes/40 moves instead of 4 minutes per game+2 second per move or 6 minutes for all the game or the same time control with ponder on instead of ponder off.
I think that usually change that does the program better at short time control does it better also at long time control and if I have the choice between one test at long time control and many tests at short time control but different type of time control then I prefer many tests at short time control.
Uri

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 18 July 2004 17:49:58:
Als Antwort auf:/In reply to: Re: sorry,, wrong button for the last message geschrieben von:/posted by: Uri Blass at 18 July 2004 17:34:43:

The question is if amateur programmers are really interested in more long time control games(there are already tournaments like Leo's tournament and the infinite loop).
I think that the time management problem may be more important and programmers may find bugs in their time management if they find that their engine perform significantly worse in 4 minutes/40 moves instead of 4 minutes per game+2 second per move or 6 minutes for all the game or the same time control with ponder on instead of ponder off.
I think that usually change that does the program better at short time control does it better also at long time control and if I have the choice between one test at long time control and many tests at short time control but different type of time control then I prefer many tests at short time control.
Uri

Hello Uri,
tests with shorter time controls -Blitz- you will always have "en masse". We do not want a short tournament with more time, but a long one, for me even a continuous one when already having a lot of games and then including updates.
It will be really possible then to see if there are engines performing better with more time than in Blitz. The quality of games is also higher and some engines need more than five or ten seconds to come across good moves.
But of course I understand that for the fast updating by doing minor changes it is more useful to have one quick test after the other. With a few exceptions like Fruit, Ufim, Thinker real progresses only happen when testing a new version once or twice a year usually - according to my experience.
Best Regards
Heinz

[Bryan Hofmann]

Geschrieben von:/Posted by: Bryan Hofmann at 18 July 2004 17:54:22:
Als Antwort auf:/In reply to: Re: sorry,, wrong button for the last message geschrieben von:/posted by: Heinz van Kempen at 18 July 2004 16:54:20:

My opinion is still the same nothing should be disabled in any engine. Stating that it is unfair that Crafty has the ability to learn and would have and advantage over other engines or it that earlier engines have it better then later engines that play against Crafty just don't hold water. The learn fuction is a integral part of Crafty and disabling it will not represent a true picture of it strength. AI is indeed the future of chess and this is but one of the examples that any other engine Author can/does use, and if they choose not to use it that's their problem.

Hello Bryan,
I understand your point of view, it was just a try to conciliate opinions here a bit. Votes so far are two votes against learning, three votes in favour of learning allowed. There are still some more votes to come and testers have time until Monday night at most to decide.
What concerns time control a majority up to now favours 40 moves in 40 minutes for each engine. A lot of different engines received points and it seems that it is not only absolute strength that counts for most testers, but also fast progresses this year, playing style and other factors, for example guessing or knowing, who might be stronger with more time.
This is all no secret for the other testers, as all received the votes from all so far.
More I do not want to tell at this time. It is a lot of suspense.
Best Regards
Heinz

Hopefully my posts were not in vain and I was able to change a few folks on this.

Most authors I have talked with are interested in long time controls as it is very time consuming for an author to preform himself.

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 18 July 2004 18:17:12:
Als Antwort auf:/In reply to: Re: Last request for testers for a common tournament geschrieben von:/posted by: Volker Boehm at 18 July 2004 15:53:04:

Hi,
are we able to send you a new version of spike for this tournament, if yes until which date? We should have enough improvements to build a Spike 0.6.
Volker

Hello Volker,
so far I only got replies in favour of allowing also devellopment versions that are still unreleased, maybe because the author (or authors in your case) want to reach a certain level before releasing. I am also in favour of allowing under the condition mentioned in the other posting.
So if there are no massive protests until tomorrows I think Spike would be a good candidate for tournament B (Higher Class) that will be a bit weaker. Those who are testing there too or exclusively will vote for the first participants on Monday/Tuesday. The exact date when we will start with tournament B might be end of this week or beginning of next week. I have to ask Volker Anuss, Thomas Mayer and Oliver Deville when they want to start with tournament B. Maybe not all the same day.
Best Regards
Heinz

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 18 July 2004 18:19:13:
Als Antwort auf:/In reply to: Re: Last request for testers for a common tournament geschrieben von:/posted by: Heinz van Kempen at 18 July 2004 18:17:12:

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 18 July 2004 18:40:26:
Als Antwort auf:/In reply to: Last request for testers for a common tournament geschrieben von:/posted by: Heinz van Kempen at 17 July 2004 17:30:38:

[Brian Richardson]

Geschrieben von:/Posted by: Brian Richardson at 18. July 2004 23:21:
Als Antwort auf:/In reply to: Re: Last request for testers for a common tournament geschrieben von:/posted by: Uri Blass at 18 July 2004 15:33:23:

Hi ,
as you can already read in those interesting discussions to the threads "just a try" and "more testers welcome" we are planning two very big tournaments for amateur engines with more time, one of them including all the strongest and fast developping engines.
The more testers participate, tha more games are played, the more interesting this tournament will be over the time. So here is your opportunity to participate in a unique common event concerning amateurs.
Any help is appreciated. Tomorrow there will be the votes, where all testers decide participants, rules, etc.. All who contribute will be able to post their own results individually, on the other hand all results also go to an overall crosstable.

Here is a small test I did with Crafty dealing with the Learn off vs on. I ran a 5/2 match Crafty 19.15 vs Riffian 1.0.5 using the Nunn I & II openings and mirroring them so both engines played each opening as White & Black. As you can see there is a 65 point difference in strength. Since there were no books used in this match this represents the strength difference of position learning only.

I am surprised to read it.
I assumed that position learning is only relevant after you get the same position that you got in previous game with position learning on so position learning can help only if Crafty has previous experience with position learning of the same position and in the nunn match crafty can have previous experience only in case that it played the same position in the past.
If I understand correctly in your experiment it played only one nunn match with positional learning on so based on my understanding positional learning could not help.
Uri

I have also been experimenting with Nunn type matches, and usually each starting position is played as both white and black, and I have found that learning is a significant factor. It should be turned off for testing, IMHO.
Brian